Migrate from NI TestStand to OpenHTF

Map NI TestStand concepts to OpenHTF — sequences and step groups to tests and phase groups, numeric limit steps to measurements, preconditions to run_if and branches, station globals to configuration, report and database loggers to output callbacks — and plan a station-by-station transition.

TestStand sequences and OpenHTF tests describe the same thing: ordered steps with limits, run against one unit, producing a report. The vocabulary differs; the shapes line up.

Concept mapping

TestStand	OpenHTF	Notes
Sequence file (`.seq`)	Python module with `htf.Test(...)`	Text, diffable, in git
MainSequence	`htf.Test(*phases)`
Setup / Main / Cleanup step groups	`htf.PhaseGroup(setup=, main=, teardown=)`	Cleanup semantics match: teardown runs if setup completed — Phase Groups
Step (Action)	Phase	A Python function
Numeric Limit Test step	`@htf.measures(htf.Measurement(...).in_range(lo, hi).with_units(...))`	GELE → `in_range`; GT/LT only → one bound; EQ → `.equals`
Multiple Numeric Limit Test	Several `Measurement`s on one phase, or a dimensioned measurement
String Value Test	`.equals("...")` or `.matches_regex(...)`
Pass/Fail Test	Boolean measurement `.equals(True)`, or `PhaseResult.FAIL_AND_CONTINUE`
Step result "Skipped"	`PhaseResult.SKIP`	Recorded as SKIP
Precondition expression	`@htf.PhaseOptions(run_if=lambda: ...)` for static, `BranchSequence` for result-dependent	`run_if` cannot see results; branches can — Checkpoints & Branching
Sequence Call	A phase list / `PhaseSequence` reused across tests; `Subtest` for an independent result	Subtests
Looping steps	`force_repeat` + `repeat_limit`, or a Python loop building phases
Step failure causes sequence failure	Default OpenHTF behaviour (continue, record FAIL); `stop_on_first_failure` or `checkpoint()` to stop	Test Options
Station Globals / Sequence File Globals	Configuration: `CONF.declare`, YAML per station
Locals / FileGlobals passed between steps	Measurements (`test.get_measurement`), plug state, `test.test_record.metadata`
Process model (serial number prompt, report, loop)	`test_start=prompt_for_test_start()`, output callbacks, `while True:` around `execute()`	Device Under Test
Operator Interface	Built-in Operator UI or TofuPilot's
Report (ATML / XML / HTML)	JSON test record via `OutputToJSON`
Database Logger	An output callback — your SQL insert, or TofuPilot's `upload()`
Code modules (LabVIEW VI, DLL, .NET)	Plug methods in Python; call DLLs via `ctypes`, LabVIEW via its Python connectivity or a rewrite	The main effort
Deployment Utility	Python packaging + a launcher — Production deployment
Batch / parallel process models	Not built in; one process per socket, or TofuPilot procedures

Worked example

A typical PCB sequence: Setup opens instruments; Main programs firmware, checks rails with numeric limits, runs an RF sub-sequence under a precondition; Cleanup powers down.

pcb01_eol.py

import openhtf as htf
from openhtf.output.callbacks import json_factory
from openhtf.plugs import user_input
from openhtf.util import checkpoints, configuration, units

from plugs import Dmm, Psu, Programmer, RfTester

CONF = configuration.CONF
RF_VARIANT = CONF.declare("rf_variant", default_value=True, description="Board has the radio fitted")


# --- Setup group: instruments open themselves in plug __init__; a phase can verify ---
@htf.plug(psu=Psu, dmm=Dmm)
def instruments_ready(test, psu, dmm):
    test.logger.info("PSU %s, DMM %s", psu.idn, dmm.idn)


# --- Main group ---
@htf.plug(prog=Programmer)
@htf.measures(htf.Measurement("fw_crc").equals("0x9A3C"))            # String Value Test
def program_firmware(test, prog):
    test.measurements.fw_crc = prog.flash("fw_2.1.4.hex")


@htf.plug(psu=Psu, dmm=Dmm)
@htf.measures(                                                          # Multiple Numeric Limit Test
    htf.Measurement("rail_3v3").in_range(3.2, 3.4).with_units(units.VOLT),
    htf.Measurement("rail_1v8").in_range(1.75, 1.85).with_units(units.VOLT),
    htf.Measurement("idle_current").in_range(maximum=0.120).with_units(units.AMPERE),
)
def check_rails(test, psu, dmm):
    psu.set(12)
    test.measurements.rail_3v3 = dmm.vdc("3V3")
    test.measurements.rail_1v8 = dmm.vdc("1V8")
    test.measurements.idle_current = psu.current()


@htf.PhaseOptions(run_if=lambda: CONF.rf_variant)                       # Precondition
@htf.plug(rf=RfTester)
@htf.measures(htf.Measurement("tx_power").in_range(17, 20).with_units(units.DECIBEL_MILLIWATTS))
def rf_tx_power(test, rf):
    test.measurements.tx_power = rf.tx_power_dbm()


# --- Cleanup group ---
@htf.plug(psu=Psu)
def power_down(test, psu):
    psu.off()


def main():
    test = htf.Test(
        htf.PhaseGroup(
            setup=[instruments_ready],
            main=[
                program_firmware,
                checkpoints.checkpoint("programmed"),    # "step failure causes sequence failure" for the expensive part
                check_rails,
                htf.Subtest("rf", rf_tx_power),          # Sequence Call with its own result
            ],
            teardown=[power_down],
        ),
        test_name="PCB01 EOL",
        sequence_version="2.1.4",                        # any FileGlobal worth keeping → metadata
    )
    test.add_output_callbacks(
        json_factory.OutputToJSON("./records/{dut_id}.{start_time_millis}.json", indent=2)
    )
    while True:                                          # process model loop
        test.execute(test_start=user_input.prompt_for_test_start())

if __name__ == "__main__":
    main()

Migrating the code modules

This is where the time goes. Options, roughly in order of preference:

Rewrite in Python against the instrument's SCPI or SDK — PyVISA, pyserial, pymodbus. Usually shorter than expected: a LabVIEW driver VI often wraps a handful of SCPI strings.
Call the existing DLL from a plug with ctypes or a vendor Python binding. Keeps validated code; adds a Windows dependency.
Keep LabVIEW for one instrument and expose it over a local TCP socket or CLI that a plug calls. A bridge, not a destination.

Transition plan

Pick one station with a simple sequence and a Python-comfortable owner. Run OpenHTF alongside TestStand on the same units for a week and compare records.
Match limits exactly from the sequence file; keep the same measurement names so historical comparisons hold.
Route both systems' results to one database so quality does not lose visibility during the switch — an output callback for OpenHTF, the existing logger for TestStand. TofuPilot accepts both OpenHTF records and imported files.
Retire TestStand licences per station as each converts; the savings fund the plug work.

Migrate from NI TestStand to OpenHTF

Concept mapping

Worked example

Migrating the code modules

Transition plan

OpenHTF vs TestStand

Phase Groups

Production deployment

On this page

Migrate from NI TestStand to OpenHTF

Concept mapping

Worked example

Migrating the code modules

Transition plan

Related

OpenHTF vs TestStand

Phase Groups

Production deployment

On this page